Gemma-4-31B-it-assistant and Gemma-4-31B-IT-NVFP4 models now available on Amazon SageMaker JumpStart

This article announces the availability of Google DeepMind's Gemma-4-31B-it-assistant and NVIDIA's Gemma-4-31B-IT-NVFP4 models on Amazon SageMaker JumpStart.

• Gemma-4-31B-it-assistant supports multimodal reasoning, coding, and agentic workflows with 256K-token context window
• Handles text, image, and video inputs with support for over 140 languages
• Ranked #3 among open models on Arena AI text leaderboard, outperforming models 20x larger
• Gemma-4-31B-IT-NVFP4 reduces memory footprint to ~18.5 GB with 4-bit FP4 quantization
• Achieves 2.5x faster inference while retaining 97-99% of original model quality
• Deploy with a few clicks via SageMaker console or Python SDK

These models enable AWS customers to deploy high-performance, scalable AI solutions optimized for different enterprise requirements.